Splitters and multipath
Last Post: January 23, 2007:
Anyone had experience here?
If I split an RF signal 4 ways and a client can see all four of these signals (although not all directly) am I opening myself up for multipath issues and black spots?
I have the most incredibly complicated automated machine to cover, where NLOS to the antenna causes dead spots. Multipath is also huge with a single antenna. There is absolutely nothing to absorb RF energy - it gets reflected/scattered like crazy.
The plan is to introduce better LOS to antennas without quadrupling the number of APS, and tune power levels down to near receiver limits (about -75dBm).
Lots of people say be careful when splitting, so is it a bad idea, or could it work?
In my opinion, there will be less multipath after you split the signal. For each split you lose at least 3dB (half) of power. In highly reflective environments it is beneficial to have less power.
You are talking about better LoS antennas. What do you mean by that? My guess is that you are using omni or sector antennas in your installation which I believe is the right way to go without knowing more about your situation. My guess is that you mean you are just going to install more antennas to get rid of the dead spots.
There are a few others on this board that I would like to see respond (M/Q) to get their input, but I think this is a viable solution as long as you are looking for a coverage model not high user density.
Good luck with it. Please post your solution and results so that will help in the future!
This sounds like a really interesting project. I have a few questions to ask as usual.
1. What frequency band are you using?
2. What size of an area do you need to cover?
3. What do you exactly mean when you say you are going to split the signal? Details please.
4. What equipment are you using now, again as much detail as you can please?
5. How are you determining that you have a multi-path problem?
6. Why is there a problem with using more and smaller RF cells?
7. Are any of the proprietary MIMO solutions a viable option, as they benefit from multi-path?
One has to be really careful when splitting a single RF source into multiple resonators. If you are not, you can actually make matters worse (been there done that). I beg your forgiveness, but I think we all would like to hear more details before we offer any suggestions.
If this is a standalone location where it is appropriate to have experimental equipment, you might want to try an Airgo third generation AP and client adapters, or just the AP.
I had good results from testing an AP with a normal (some might say subnormal) Intel 2200BG radio on a tablet computer in an office building. This probably didn't provide loads of bad multipath, but initial results were superior to a standard enterprise AP in the same location.
If all of the clients will be within 10-15 feet of where leaky coax could be run, it has a very controlled field. It would require an FCC experimental permit, I think, and I don't know that it would support 54 Mbps, but it might work ok for 802.11b speeds.
Thanks for the replies. Unfortunately, I can't give too much detail about the machine, but I will try to paint a better picture of the problem.
This is a 2.4GHz low bandwidth 802.11b application - there are approx 180 cars with 11b boards fited with dipoles which move in a fixed horizontal plane, each on their own track of about 100ft. These tracks are stacked into an array of 12w x 14h. The total width of the machine is about 100ft.
The initial solution is to use 12 directional antennas at the end of the machine, but the gain is too high with the current system and the environment is reflective with high delay-spread multipath. I have been able to identify this with a yellowjacket analyser, which also shows me a lot of RF nulls caused by metal struts in the machine.
The beam pattern of directional antennas we have found has been far too wide giving the signal far too much scope to reflect off the metal surfaces and even the warehouse walls and ceiling. Reducing antenna gain affects the ability of the cars to receive at the far ends as there are mesh floors for access which scatter the direct signal.
Power levels are measured at a about -50 to -60dBm at most places in the machine, but link quality is poor.
In order to cover with direct LoS, we would need 48 APs and antennas. I don't want to put this many APs in :o This is why I want to use 4-way splitters in order to keep the number of APs down to 12.
I have been playing with a single run of leaky co-ax which runs the length of the track, and results are good for multipath, and the signal strength is surprisingly good over the run (1mW transmit power at the AP yields a -50dBm signal at the receiver). I'd like to try 4 runs of this in adjacent aisles with a 4-way splitter (6dB loss) from a single AP, but it could be an expensive way of finding out this solution is no better. This is my main concern. I predict that I'll be engineering in some multipath by doing this, and there could be areas of destructive interference. I would guess the delay spread should be a lot shorter from an identical signal on the co-ax next door than from a bounce off the wall or roof, but I am just hypothesising - the warehouse environment has proven my theories to be wrong time and time again.
Thanks for the explanation. It really does sounds like a very interesting project. I did not realize you where using radiating coax. This link maybe of some interest to you.
I think I would try another run of coax. Interference problems are specific to each individual application. I still would like to know how you intend to split the signal path as that is important. Any slight difference in feeder length to the coax will have an affect. The strange thing is what ever that affect might be could either hurt or help you. The only way to know is to test it.
It is a great that you have the YellowJacket to help you, without that it would be quite a bit more difficult.
Thanks Michael - this is the conclusion that I am coming to. I will just have to test and see what happens. I don't know what I would do without my yellowjacket!
I am planning to split with a Huber and Suhner power splitter type 5502.19.0006 - see http://www.hubersuhner.com/hs-p-rf-ps-td.pdf . This is a passive device that will combine the feeders into the AP antenna.
The feeder cables will be exactly the same length.
That appliance is exactly what I would have used as well.
One final thought and that is if you are seeing problems, varying the coaxial length so that it is different between the two radiators might help. There is some science to it, changing the impedance will affect the timing and that might make things better. That is of course if the RF conditions stay relatively the same, and that appears to be the case in this situation.
I sincerely hope that you have enough time to keep us abreast of your research, I would be really interested to hear your results.
Personally I'd be reluctant to use splitters and a leaky feeder on such a scale because you can have all manner of problems when using splitters and antennae in this way.
Equally, you need to be careful when using splitters with normal antennae. If there is too much seperation between the antennae, or if you have vastly different cable lengths from the splitter to each antenna, it will confuse the hell out of your AP.
What you're effectively talking about is implementing antenna diversity (the splitter and the antenna/leaky-feeder) but without actually having a radio that supports diversity. I think you'd be better off looking at either;
(A) Using a MIMO type implementation as already suggested
(B) Use an 802.11A/B/G AP that supports diversity antennae
MIMO isn't an option here
The APs I am using are capable of diversity, but that's not what I am trying to achieve. I am trying to get better LoS to the client devices without quadrupling the number of APs I am putting in. This is because the RF picture in the machine will continually be in a state of flux, and installing a NLoS solution isn't something that I would relish when I am being called at 4am because of a problem.
My understanding of antenna diversity is that it provides a compatible receiver with another 'opinion' of the received signal, such that it can decide which has the best signal strength prior to decoding. If I were to implement this, then wouldn't I need 2 leaky feeders down each track (about a wavelength apart)? If I were to use the diversity mode antenna connections on my AP to go to two different tracks, wouldn't the coverage areas be significantly different enough to cause even greater problems?
I was always under the impression that diversity should not be used to provide RF cover in two different areas, which is why I have proposed a splitter-based solution?